BANCO CENTRAL DE RESERVA DEL PERÚ Habit formation and sunspots in overlapping generations models Fabrizio Orrego* * Carnegie Mellon University and Central Bank of Peru. DT. N° 2011-013 Serie de Documentos de Trabajo Working Paper series Agosto 2011 Los puntos de vista expresados en este documento de trabajo corresponden al autor y no reflejan necesariamente la posición del Banco Central de Reserva del Perú. The views expressed in this paper are those of the author and do not reflect necessarily the position of the Central Reserve Bank of Peru. Habit formation and sunspots in overlapping generations models Fabrizio Orregoy Carnegie Mellon University Central Bank of Peru First version: August 2009 This version: July 2011 Abstract I introduce habit formation into an otherwise standard overlapping generations econ- omy with pure exchange populated by three-period-lived agents. Habits are modeled in such a way that current consumption increases the marginal utility of future consumption. With logarithmic utility functions, I demonstrate that habit formation may give rise to stable monetary steady states in economies with hump-shaped endowment pro…les and reasonably high discount factors. Intuitively, habits imply adjacent complementarity in consumption, which in turn helps explain why income e¤ects are su¢ ciently strong in spite of logarithmic utility. The longer horizon further strengthens the income e¤ect. Finally, I use the bootstrap method to construct stationary sunspot equilibria for those economies in which the steady state is locally stable. I thank Stephen Spear, Laurence Ales and Yaroslav Kryukov for their support and detailed advice through- out this project. I am also grateful to two anonymous referees for many useful comments and suggestions on this paper, and seminar participants at Carnegie Mellon University, the University of Pittsburgh and the Central Bank of Peru. As usual, all remaining errors are my own. yEmail address: [email protected] 1 1 Introduction In textbook models of overlapping generations (OLG), sunspot equilibria have been intimately related to backward bending o¤er curves and stable monetary steady states (see Blanchard and Fischer [6]). Theoretically, if income e¤ects are strong enough, then aggregate savings is a strictly decreasing function of the real return and the monetary steady state is a sink. In two- period models equipped with power utility, such income e¤ects require arguably unrealistic paramenter speci…cations (discount factor close to zero, income concentrated in the …rst period of life and high curvature of utility function), which may explain the apparent reluctance of the profession to embrace the idea of a sunspot equilibrium. In this paper I introduce habit formation into an otherwise standard three-period-lived OLG economy, in an attempt to overcome the legitimate concern discussed above. I show that the steady state is stable when the discount factor is close to one-half, the endowment pro…le is hump-shaped and the utility function is logarithmic. Then I use the forward stability of the steady-state dynamics to construct local sunspot equilibria around the steady state, as in Spear, Srivastava and Woodford [24]. In light of recent economic developments, the fresh perspective on stability presented in this paper can also provide an alternative explanation, via the sunspot equilibria, for the excess volatility that is currently observed, or for the fact that the great moderation in macroeconomic variability in the U.S. may well be part of an on-going cyclic equilibrium variation1. I assume that agents derive utility from the gap between current and past consumption. In presence of subtractive habits not only the elasticity of intertemporal substitution is less than one at any steady state, but also dated consumption goods are weak gross complements at yields near the golden rule. I show that the set of economies in which the steady state is a sink is relatively large when the discount factor is close to zero, but shrinks gradually as the discount factor approaches one-half. Intuitively, when the discount factor is small enough, the desire for youth consumption increases, which a¤ects positively the marginal utility of consumption in the second stage and fosters adult consumption. This in turn increases the marginal utility of consumption in the third stage and so on. The adjacent complementarity in consumption helps explain why income e¤ects are su¢ ciently strong in spite of the logarithmic utility. On the other hand, the longer horizon further strengthens the income e¤ect and helps support stable steady states for reasonable values of the discount factor. The intuition is as follows. When the monetary steady state is stable, the equilibrium sequence of returns converges to the golden rule via damped oscillations. Note that in the three-period case, households demand assets when young and middle-aged. If returns are high today, then the current middle-age agents’wealth is small, since they received low returns on their savings 1 I thank Stephen Spear for this idea. 2 yesterday. Consequently, their demand for money is low. This in turn suggests that, even if the young agents’ demand for money is high because of the high returns, the aggregate demand for savings can be low because the middle-age agents are constrained by their wealth. Thus, because of the wealth distribution among di¤erent generations, the three-period model possesses a mechanism for causing aggregate asset demand to decrease with the current rate of return that is completely absent from the two-period model. The remainder of the paper is structured as follows. Section 2 introduces the layout of the model. Section 3 derives the competitive equilibrium. Section 4 deals with the competitive equilibrium with sunspots. The last section concludes. Related literature Under the condition that the monetary steady state is stable, there can be a continuum of convergent solutions, as well as cyclic solutions. In the former situation, I say that the competitive equilibrium is indeterminate2. Kehoe and Levine [18] show robust examples of indeterminacy in three-period-lived OLG economies with outside money if the discount factor is one-half, income is hump-shaped and the intertemporal elasticity of substitution is one fourth3. An important property of an intertemporal elasticity of substitution of one-fourth is that it allows goods in di¤erent periods to be gross complements at some prices. In situations where all goods are gross substitutes, Kehoe, Levine, Mas-Colell and Woodford [19] show that indeterminacy of the type discussed above is impossible. Theoretical foundations for several versions of intrinsic habit formation have been recently provided by Rozen [22]. In the OLG framework, habit formation creates room for indetermi- nacy and is open to a …nite number of di¤erent speci…cations, with diverse implications on asset prices. To the best of my knowledge, Lahiri and Puhakka [20] …rst incorporate subtrac- tive habit formation in a two-period OLG model with outside money. They show that cycles arise when the discount factor is near zero and utility is logarithmic, because generational o¤er curves must bend backwards (however, later on in this paper I establish that the o¤er curve does not bend backwards in their model). The current three-period extension of Lahiri and Puhakka [20] is by no means trivial, as several authors have demonstrated that an economy in which the life cycle goes to three 2 On the other hand, Bhattacharya and Russell [5] show that in three-period OLG economies there exist two-period cycles if the elasticity of substitution is less than one-fourth and income is positive only in the …rst period of life, for any value of the discount factor. 3 Unfortunately, it remains di¢ cult to support indeterminate equilibria in multiperiod OLG economies with pure exchange, outside money and more complex structures. For instance, Azariadis and Lambertini [2] …nd that indeterminacy arises in three-period economies in the presence of endogenous borrowing constraints, logarithmic utility functions and hump-shaped endowment pro…les. However, their result is not robust to the introduction of unproductive assets in positive net supply. 3 periods or more may behave di¤erently from the well-known Diamond model (see for example Azariadis, Lee and Ohanian [3], Kehoe and Levine [18] and Henriksen and Spear [16]). I use in this paper the bootstrap method studied by Spear, Srivastava and Woodford [24] to construct stationary rational expectations equilibria with sunspots, which are known to exist in OLG models according to the Philadelphia Pholk Theorem4. In the derivation of these equilibria, I …nd that prices depend on lagged endogenous variables, as in Henriksen and Spear [16]. This result was …rst presented by Spear, Srivastava and Woodford [24] in a two-period model with n > 1 goods and m > 1 agents. The similarity is not surprising, because the three-period-lived model may also be interpreted as a two-good, two-consumer, two-period-lived model (see Balasko, Cass and Shell [4] and Kehoe and Levine [18]). In this paper I stick to the former speci…cation since it drastically reduces the number of parameters that are not subject to normalization and therefore the analysis remains as simple as possible. 2 The model I study a closed, pure-exchange economy of overlapping generations of agents who live for three periods, labeled as young adulthood, middle age and retirement. There is one individual in each cohort and no population growth. Time is discrete. In period 1, in addition to the young consumer, there are two others, an old consumer who lives in period 1 only and a middle-aged consumer who lives in periods 1 and 2. Agents are allocated exogenous amounts of the unique consumption good in each period of life. I assume, as in Geanakoplos, Magill and Quinzii [14], that these endowments are stationary, and the typical young agent’slifetime endowment 3 is ! = (!1;!2; 0) R+, which can be interpreted as the agent’s labor income (income in 2 retirement is zero).